Thanks BreMaster, but I guess I still don't get it completely. For newly mined blocks I understand why the aggregation of 51 percent of hashing power helps the attacker.
If the attacker decides to fork much earlier, let's say 2,500 blocks earlier, he has to catch up and eventually overtake the main chain. The attacker cannot do so however, because the difficulty must be adjusted if he mines blocks too fast. So, even if 2016 blocks are mined every 2 weeks on average, this applies to both chains. So rather than catching up with the main chain, the attacker will fight a loosing battle, since the other chain eventually grows as fast as his.
I know this is only a theoretical problem, but I was wondering whether the blocktime itself (or how you want to call it) imposes limitations to the 51 percent attack. Thanks!